位置=“闪避”不闪避

Question

我正在绘制R中的两个条形图，其中我对其中一个条具有置信区间，并且遇到了一个问题，即position =“ dodge”或position = position_dodge（）根本不会移动这些条。我查看了关于stackoverflow的类似问题，但似乎不多。

我的代码是：

ggplot(data = Estimates.df, aes(x = Assets, y = Estimate, group = Estimate)) +
  geom_bar(data = Estimates.df, aes(x = Assets, y = Estimate, group = Estimate, fill = Estimate), stat = "identity", fill = "Blue", position = "dodge") +
  geom_bar(data = Estimates.df, aes(x = Assets, y = Target, group = Target, fill = Target), stat = "identity", fill = "Black", position = "dodge") +
  coord_flip() +
  geom_errorbar(aes(ymin = (Estimate - 1.645*Std..Error), ymax = (Estimate + 1.645*Std..Error), group = Estimate), position = "dodge")

我需要估计图上的置信区间，而不是目标图上的置信区间。我尝试通过融化将数据帧转换为long，但是无法以这种方式保持置信区间。

谁能看到我哪里出了问题？

我的数据是：

Assets <- c("TW", "Asset1", "Asset2", "Asset3", "Asset4", "Asset5", "Asset6", "Asset7", "Asset8", "Asset9", "Asset10", "Asset11", "Asset12", "Asset13", "Asset14", "Asset15", "Asset16")
Std..Error <- c(.0002,.050,.0147,.028,.0289,.035,.0184,.0709,.0130,.0356,.0071,.0067,.0080,.0162,.0188,.0099,.0147)
Estimate <- c(.002,-0.032,.0317,.018,.0336,.0426,.09774,.3191,.0067,.0892,.0225,-0.0145,.0263,.01125,.07359,.09055,.1843)
Target <- c(0,.05,0,.04,0,0,.07,.23,0,.18,.05,0,.02,0,.09,.05,.22)

Estimates.df <- data.frame(cbind(Assets, Std..Error, Estimate, Target))

Answer 1

您的代码有几个错误，您需要ggplot2的长格式数据（不是宽格式），因此我将tidyr和dplyr的数据转换为长格式。然后，您不应在ggplot的每个参数中重复美学。最后，您需要确保值的类型为numeric，因为它们的类型为factor。

Estimates.df <- data.frame(cbind(Assets, Std..Error, Estimate, Target))
library(tidyr)
library(dplyr)
Estimates.df<-Estimates.df %>%
  gather(type,value,Estimate:Target)


ggplot(data = Estimates.df, aes(x = Assets, y = value, fill=type, color=type)) +
  geom_bar(stat = "identity",
           position = "dodge")+
  coord_flip() +
  geom_errorbar(
    aes(ymin = (as.numeric(value) - 1.645*as.numeric(Std..Error)),
        ymax = (as.numeric(value) + 1.645*as.numeric(Std..Error))),
    position = "dodge")

Answer 2

您将需要使用长数据。我使用melt来做到这一点，并在绘制之前弄乱了结果，这样就不会像您所说的那样获得Target的误差线。我也更喜欢在之前创建ymin和ymax。 以下是我认为您要求的颜色以外的颜色。

library(reshape2)
Estimates.df <- data.frame("Assets" = c("TW", "Asset1", "Asset2", "Asset3", "Asset4", "Asset5", "Asset6", "Asset7", "Asset8", "Asset9", "Asset10", "Asset11", "Asset12", "Asset13", "Asset14", "Asset15", "Asset16"),
                           "Std..Error" = c(.0002,.050,.0147,.028,.0289,.035,.0184,.0709,.0130,.0356,.0071,.0067,.0080,.0162,.0188,.0099,.0147),
                           "Estimate" = c(.002,-0.032,.0317,.018,.0336,.0426,.09774,.3191,.0067,.0892,.0225,-0.0145,.0263,.01125,.07359,.09055,.1843),
                           "Target" = c(0,.05,0,.04,0,0,.07,.23,0,.18,.05,0,.02,0,.09,.05,.22))

Estimates.df.p <- Estimates.df %>% 
  mutate(ymin = Estimate - 1.645*Std..Error, ymax = Estimate + 1.645*Std..Error) %>% 
  melt(id.vars = c("Assets", "ymin", "ymax", "Std..Error"), value.name = "value", variable.name = "variable") %>% 
  mutate(ymin = ifelse(variable == "Target", NA, ymin),
         ymax = ifelse(variable == "Target", NA, ymax))


ggplot(data = Estimates.df.p, aes(x = Assets, y = value, group = variable)) +
  geom_bar(aes(fill = variable, colour = variable), stat = "identity", position = "dodge") +
  coord_flip() +
  geom_errorbar(aes(ymin = ymin, ymax = ymax, group = variable), position = "dodge")